Using a combined machine learning and bio-automation approach to understand, recognise and control the bacterial stress landscape

这项研究旨在更好地了解细菌在经历压力时所经历的调节变化，以及如何利用这些变化来生产细菌压力报告菌株。具体而言，研究集中在负荷应激的概念：产生大量给定产物的代谢消耗，例如高度表达的蛋白质。本项目选择的微生物是大肠杆菌，大肠杆菌在工业中广泛使用，被认为是革兰氏阴性菌的模式生物，因此具有很好的特征，经常用于研究。对细菌压力景观的研究仍然相对新颖，因此，该项目预计将产生影响，有助于进一步研究，特别是在合成生物学和生物技术行业。此外，这些知识可以应用于工业环境中，如果更好地了解负载应力的影响，可以提高产量和效率。该研究旨在识别，响应和定义包括负载应力在内的应力响应的监管指纹。为了实现这一目标，将需要一个系统，可以在细菌中诱导负荷应力，以及一个系统，以监测和测量的监管和生理影响应力在全球和单细胞水平。在细菌中诱导负荷应激将通过引入成对的合成构建体来实现，所述合成构建体（i）促进一系列异源蛋白的表达，（ii）使用合成和天然启动子监测不同异源蛋白的表达对细胞的调控结构的影响。机器学习和优化算法将有助于实验设计，而生物自动化将允许更大规模的测量和最终的假设检验。结果的分析将需要大量的信息学，以探索从两个组件系统到调节子和启动子架构的不同响应水平的调控网络。它的目的是在这些领域开发新的方法，可以推广到本研究以外的更广泛的应用。最终，这项研究的结果可以帮助生物技术行业提高有用化合物的生产效率，并帮助我们更多地了解细菌对压力的反应方式。

This research aims to better understand the regulatory changes that bacteria undergo whilst experiencing stress and how to harness these changes for producing bacterial stress reporter strains. Specifically, the research is focused the concept of load-stress: the metabolic exertion of producing large amounts of a given product e.g. a highly expressed protein. The chosen organism for this project is Escherichia coli which is widely used in industry and is considered to be the model organism for gram-negative bacteria, as such is very well characterised and is frequently used in research. Investigations into the stress landscape of bacteria is still relatively novel, therefore, this project is expected to have an impact that will assist with further research especially in synthetic biology and the biotechnology industries. Additionally, this knowledge can be applied in an industrial setting where yield and efficiency can be improved given a better understanding of the impact of load-stress.The research will aim to recognise, respond to, and define the regulatory fingerprints of stress responses including load stress. To achieve this aim, a system will be required that can induce load-stress in bacteria as well as a system to monitor and measure the regulatory and physiological impacts stress at a global and single cell level. Inducing load stress in bacteria will be accomplished by introducing paired synthetic constructs (i) that promote expression of a range of heterologous proteins (ii) monitor the effects of expression of differing heterologous on the regulatory architecture of the cell using synthetic and natural promoters. Machine learning and optimisation algorithms will aid the design of experiments while bio-automation will allow a greater scale of measurement and ultimately hypothesis testing. The analysis of results will require a significant amount of informatics in order to explore the regulatory networks at different levels of response from two components systems to regulons and promoter architectures. It is intended to develop new methodology in these areas that can be generalised for wider application outside of this study. Ultimately, the outcome of this research could help the biotechnology industry increase the efficiency of the production of useful compounds and will help us understand more about the way bacteria respond to stress.